Auxiliary diagnosis method and system based on depth learning

Abstract

The invention discloses an auxiliary diagnosis method based on depth learning. The method comprises steps that S1, segmentation for raw corpus data is carried out to establish a word embedding query table; S2, a training set is generated based on key characteristic fields of electronic case data, the word embedding query table is utilized for digitalization, the convolutional neural network is utilized to generate an auxiliary diagnosis model; and S3, key characteristic fields of a newly-inputted electronic case are extracted, digital conversion is carried out through the word embedding querytable, the auxiliary diagnosis model is utilized for matching, and the matched diagnosis result is outputted. The invention further provides an auxiliary diagnosis system based on depth learning. Thesystem comprises a corpus data extraction module, a word embedding query table construction module, a historical electronic case data extraction module, a new electronic case data extraction module, asegmentation module, an electronic case digital module and an auxiliary diagnosis module. The method is advantaged in that the diagnosis result is timely and accurate, and doctors are effectively facilitated to carry out rapid condition diagnosis.

Description

technical field [0001] The invention relates to the field of big data analysis, in particular to an auxiliary diagnosis method and system based on deep learning. Background technique [0002] Usually, doctors in large public hospitals need to see a large number of patients with similar symptoms every day, including inexperienced young doctors. When there are too many patients, it is easy to have problems such as low doctor work efficiency and even high misdiagnosis rate. [0003] With the advancement of my country's health and medical big data planning, the hospital's medical record system has entered the information age, and a large amount of electronic medical record data has been accumulated in the hospital. These electronic medical record data contain detailed records of patients in hospital diagnosis, including symptoms, illness and treatment measures, etc., which are of high reference value for doctors to make a diagnosis. [0004] In recent years, deep learning has d...

AI Technical Summary

Problems solved by technology

Considering that electronic medical records have special fields in the data (such as current medical history, physical examination results), they have high reference value for auxiliary diagnosis. For this, traditional auxiliary diagnosis methods or systems rely on extremely standardized data. During the standardization process The overhead is huge, and because the data is too standardized, it is impossible to identify whether the "cough for a week" and "cough for a week" (both are conventional notations) input by different doctors in practical applications are the same attribute; at the same time, the traditional auxiliary diagnosis method Or when the system extracts medical record features, it extracts individual words instead of the features of the entire sentence. When there are many samples, there are too many features, and complete extraction is impractical and not suitable for large-scale data.

Images